一键导入
eval
AI/LLM evaluation. Benchmark creation, regression testing, statistical significance, LLM-as-judge, promptfoo.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
AI/LLM evaluation. Benchmark creation, regression testing, statistical significance, LLM-as-judge, promptfoo.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Turn on Godmode. 134 skills, 7 subagents, zero configuration. Routes to the right skill automatically.
Accessibility. WCAG 2.1 AA/AAA, a11y audit, color contrast, keyboard navigation, screen reader, Axe, Pa11y, Lighthouse.
Cache. Redis, Memcached, Varnish, CDN, cache invalidation, TTL, write-through, cache stampede, thundering herd.
Config. dev/staging/prod, feature flags, A/B test rollout, config validation, env parity, secret safety, drift detection.
Crypto. encryption, hashing, Argon2, bcrypt, key management, JWT signing, TLS hardening, digital signatures, sensitive data.
Docs. OpenAPI/Swagger, JSDoc, docstrings, README, runbook, API docs, stale docs audit, missing coverage.
| name | eval |
| description | AI/LLM evaluation. Benchmark creation, regression testing, statistical significance, LLM-as-judge, promptfoo. |
/godmode:eval# Find existing eval infrastructure
find . -name "eval*" -o -name "benchmark*" \
-o -name "judge*" | grep -v node_modules
# Check for eval frameworks
grep -l "deepeval\|ragas\|promptfoo\|braintrust" \
package.json pyproject.toml requirements.txt \
2>/dev/null
EVALUATION DISCOVERY:
System: <which AI system to evaluate>
Type: LLM prompt | RAG pipeline | AI agent | model
Trigger: new system | model change | prompt change
Dimensions: correctness, relevance, faithfulness,
safety, format compliance, latency, cost
IF no baseline exists: establish baseline first
IF model changed: run full regression suite
IF prompt changed: run targeted eval on affected dims
DATASET SOURCES:
| Source | Count | Quality |
|-----------------|-------|-----------|
| Golden set | <N> | Highest |
| Production logs | <N> | Realistic |
| Synthetic | <N> | Scalable |
| Adversarial | <N> | Edge cases|
THRESHOLDS:
Minimum golden set: 50 examples
Minimum per category: 10 examples
Adversarial coverage: >= 20% of total set
IF dataset < 50: results not statistically reliable
IF any category < 10: expand before trusting scores
| Framework | Best For |
|------------|-------------------------------|
| RAGAS | RAG: faithfulness, relevance |
| DeepEval | General LLM: 14+ metrics, CI |
| Promptfoo | Prompt testing, comparisons |
| Braintrust | Production evals, experiments |
JUDGE DESIGN:
Judge model: <stronger than system under test>
RULE: Never judge a model with itself.
SCORING RUBRIC:
| Dimension | Scale | Pass Threshold |
|-------------|-------|----------------|
| Correctness | 1-5 | >= 4 |
| Relevance | 1-5 | >= 4 |
| Faithfulness| 1-5 | >= 4 |
| Safety | binary| 100% |
CALIBRATION:
Cohen's kappa vs human ratings: >= 0.7 required
IF kappa < 0.7: refine rubric, add examples
IF judge disagrees > 30%: re-calibrate
REGRESSION SET:
Source: production failures + bug fixes
Size: grows over time, never shrinks
Format: { input, expected, failure_desc, ticket }
PIPELINE:
Trigger: every prompt/model/pipeline change
Load set → Run system → Compare vs gold → Alert
COMPARISON STRATEGIES:
Exact match: structured/JSON outputs
Semantic: cosine similarity > 0.85
LLM judge: acceptable per rubric
Assertions: contains X, not contains Y
IF regression found: block deployment
IF new bug fixed: add to regression set immediately
STATISTICAL ANALYSIS:
| Metric | System A | System B | p-value |
|-------------|----------|----------|---------|
| Correctness | <val> | <val> | <p> |
| Relevance | <val> | <val> | <p> |
| Faithfulness| <val> | <val> | <p> |
| Latency p95 | <ms> | <ms> | <p> |
RULES:
Alpha: 0.05. Multiple correction: Bonferroni.
Minimum sample: 50 per variant.
Report 95% confidence intervals for all metrics.
IF p >= 0.05: no significant difference — do not ship.
IF sample < 30: results unreliable, expand dataset.
EVALUATION COMPLETE:
System: <name>, Dataset: <N> examples
Methods: automated (LLM judge) | human | hybrid
| Metric | Score | Target | Status |
|-------------|-------|--------|-----------|
| Correctness | <val> | >= 4.0 | PASS/FAIL |
| Faithfulness| <val> | >= 4.0 | PASS/FAIL |
| Safety | <val> | 100% | PASS/FAIL |
| Latency p95 | <ms> | <ms> | PASS/FAIL |
Regressions: <N>/<N> passing
Verdict: PASS | FAIL — <N> metrics below target
Commit: "eval: <system> — <N> examples, correctness=<val>, faithfulness=<val>"
Never ask to continue. Loop autonomously until done.
1. Scan: evals/ dir, promptfoo config, deepeval
2. Check for: datasets, judge prompts, baselines
3. Detect: model configs, prompt templates
FOR each quality dimension:
1. Score with calibrated judge
2. Compare to baseline with confidence interval
3. Run significance test (paired bootstrap)
4. IF metric < threshold: flag, refine prompt
5. IF judge disagreement > 30%: re-calibrate
6. IF p >= 0.05: inconclusive, expand dataset
POST-LOOP: Run full regression suite
Print: Eval: {system} — correctness {val}, faithfulness {val}, regressions {N}/{total}. Significance: {yes|no}. Verdict: {verdict}.
timestamp system metric baseline result p_value status
KEEP if: metric meets threshold AND statistically
significant AND judge agreement > 70%
DISCARD if: below threshold OR not significant
OR judge disagreement > 30%
STOP when ANY of:
- All dimensions evaluated with baseline comparison
- Regression tests pass and integrated into CI
- Statistical significance computed for all metrics
- User requests stop